Mpirun problem?

[dquan]

I was trying to run a hden grid from 1.7 to 9.7 in 0.5 dex step. T is fixed to 100K. I used mpirun and found the following three grid points (hden=4.2, 4.7, 5.2) were missing in .grd file. Gary ran the model in serial and the program aborted at hden=4.2 due to a negative density.

Strange thing is that although the mpirun main output file says at the three settings there is an error due to the negative density of NH3, there is no abortion reported and the program just went to the next setting.

This input ran ok in newmole branch (version r3949).

I attached the input file.

[gary]

the problem is that the abort was not reported in the "save grid" output. The negative populations problem does not occur with newmole so is not something we need to work on. But the "save grid" command not properly reported aborts in mpi mode is a problem.